The present invention relates to a laser processing method and a laser processing apparatus, and particularly, to an improvement of a throughput in an optical annealing step with respect to the manufacture of an insulated gate type semiconductor device such as a thin-film transistor (TFT) which is formed on a substrate having a crystalline silicon film of non-single crystal and other semiconductor devices.
Particularly, the present invention relates to the manufacture of a semiconductor device which is formed on an insulating substrate made of glass or the like and having a large area.
Recently, studies have been made of an insulated gate type semiconductor device including a thin-film like active layer (also referred to as active region) on an insulating substrate. Particularly, studies have been made earnestly of thin-film like gate transistors, so-called thin-film transistors (TFT). These transistors are classified by the material and crystalline state of a used semiconductor into an amorphous silicon TFT, a crystalline silicon TFT, and the like. The crystalline silicon is not a single crystal but a non-single crystal. Accordingly, the general term for these transistors is a non-single crystal TFT.
In general, the mobility of an amorphous semiconductor is small, so that it can not be used as a TFT which is required high sped operation. Further, since the mobility of a P-type amorphous silicon is extremely small, thereby being unable to manufacture a P-channel TFT (TFT of PMOS) so that it is impossible to form a complementary MOS circuit (CMOS) by combining the P-channel TFT with an N-channel TFT (TFT of NMOS).
On the other hand, the crystalline semiconductor has a mobility larger than that of the amorphous semiconductor, so that high speed operation can be achieved. By the crystalline silicon, not only TFT of NMOS but also TFT of PMOS can be obtained, so that it is possible to form a CMOS circuit.
The crystalline silicon film of non-single crystal has been obtained by thermally annealing an amorphous silicon film obtained by a vapor phase deposition method for a long time at an appropriate temperature (normally more than 600.degree. C.) or by irradiating it with the intense light such as a laser beam (optical annealing).
However, in the case where a glass substrate which is cheap and rich in workability, is used as an insulating substrate, it has been extremely difficult to obtain the crystalline silicon having a sufficiently high mobility (so high that a CMOS circuit can be formed) by only the thermal annealing. This is because the above-mentioned glass substrate has generally a low distortion point temperature (about 600.degree. C.), so that it is impossible to increase the substrate temperature up to a temperature required to form the crystalline silicon film having the sufficiently high mobility.
On the other hand, in the case where the optical annealing is used to crystallize a silicon film based on a glass substrate, it is possible to give a high energy to only the silicon film without increasing the substrate temperature to a very high temperature. Thus, the optical annealing technique is regarded as very effective for crystallizing the silicon film based on the glass substrate.
At present, a high power pulse laser such as an excimer laser is most preferable as an optical source for the optical annealing. The maximum energy of this laser is very large as compared with a continuous-wave laser such as an argon ion laser, so that it has been possible to improve the throughput by using a large spot of more than several cm.sup.2. However, when a normally used square or rectangular beam is used, it must be moved up and down and right and left to process one substrate having a large area. Thus, there is a room for improvement from the viewpoint of the throughput.
Concerning this, much improvement has been obtained by transforming a beam into a liner beam to extend the length of the beam (largeness of the cross section of the linear beam in the longitudinal direction) over a substrate to be processed, and by moving this beam relatively to the substrate to scan.
Here, the scanning means that irradiation of the laser beam is performed while the linear laser beam is moved in the line width direction (direction orthogonal to the longitudinal direction of the cross section of the linear beam), and the irradiated regions are overlapped with each other not to separate the irradiated regions. Also, in general, when the irradiation of the linear laser beam is performed for a large area, the scanning paths are made parallel to each other.
Further, before the optical annealing, when the thermal annealing is carried out, it is possible to form a silicon film having more superior crystallinity. With respect to the method of the thermal annealing, as disclosed in Japanese Patent Unexamined Publication No. Hei. 6-244104, by using the effect that an element such as nickel, iron, cobalt, platinum, or palladium (hereinafter referred to as crystallization catalytic element or simply referred to as catalytic element) accelerates the crystallization of amorphous silicon, the crystalline silicon film can be obtained by the thermal annealing at a lower temperature for a shorter time than a normal case.
However, in the above irradiation of the linear laser, in relation to the maximum energy thereof, the length of the linear laser beam (largeness of the cross section of the laser beam in the longitudinal direction) has been limited to about 20 cm at best.
If the processing is performed by the linear laser beam having a length longer than the limit, the energy density of the laser beam becomes insufficient to, for example, crystallize the amorphous silicon film. Thus, when a substrate having a large area is used and laser processing is performed for a region longer than the length of a linear laser beam, it has been necessary to perform scanning of the laser beam up and down and left and right, that is, both in the line width direction and in the longitudinal direction. FIG. 13(B) schematically shows scanning paths of a conventional laser beam.
FIG. 13(A) is a sectional view of a linear laser beam, and FIG. 13(B) is a view showing a surface to be irradiated viewed from the above. As shown in FIG. 13(A), an end portion la of a linear laser beam 1 is not completely rectangular, and the energy density in this portion is dispersed.
As shown in FIG. 13(B), the scanning of the linear laser beam 1 is performed along two scanning paths 2 and 3. For example, after the downward scanning of the linear laser beam 1 is performed along the left scanning path 2, the downward scanning is performed along the right scanning path 3. At this time, it is necessary to perform scanning so that the end portions 1a of the linear laser beams 1 are overlapped with each other. Then, it becomes a problem how to overlap the end portions la of the linear laser beams 1. In FIG. 13(B), a region 4 shown in a rectangle is a region where scanning is performed by the overlapped end portions la of the linear laser beams 1 in the surface to be irradiated.
However, in general, since it is difficult to control the energy density at the end portion 1a of the linear laser beam 1, semiconductor devices formed in the region 4 and in the neighborhood thereof are of extremely uneven characteristics as compared with devices formed in other region. Thus, the semiconductor material in the region 4 is not suitable for processing of semiconductor devices.
As a countermeasure to the above problem, by irradiation of a laser beam through a slit, the end portion in the longitudinal direction in which the control of energy density is difficult, is shielded to shape the end portion of the laser beam. FIG. 14(A) is a sectional view showing a linear laser beam shaped by the slit, and FIG. 14(B) is a schematic view showing scanning paths of the laser beam and is a view showing a surface to be irradiated viewed from the above.
As shown in FIG. 14(A), through the slit, an end portion 5a of a laser beam 5 is shaped into a rectangle, so that the distribution of the energy density in the end portion 5a becomes uniform than the linear laser beam 1 shown in FIG. 13(A). As shown in FIG. 14(B), when the irradiation of the linear laser beam 5 is performed, for example, the following scanning steps may be made: after the downward scanning of the linear laser beam 5 is performed along a left scanning path 6, the downward scanning is performed along a right scanning path 7. At this time, the scanning is performed so that the end portions 5a of the linear laser beams 5 are overlapped with each other. However, since the end portion 5a of the laser beam 5 is shaped into a rectangle, and the energy density distribution is uniform, it is sufficient to overlap the end portions 5a of the linear laser beam 5 with each other to the extent that the end portions 5a are brought into contact with each other, as shown by reference numeral 8. Thus, it is possible to reduce the region 8 where the ends 5a are overlapped with each other.
However, even if the energy density in the end portion 5a of the laser beam 5 is controlled by using the slit, the semiconductor devices formed in the region 8 to be scanned with the overlapped end portions 5a of the laser beam 5, are of remarkably uneven characteristics as compared with devices formed in other region.